Unexpected transformations of an azoxyquinoxaline

Treatment of N , N ’-di(quinoxalin-2-yl)diazene N -oxide 3 with strong acids did not give the expected Wallach-type hydroxylated product, but the first representative of the pentacyclic imidazo[1,2-a :4,5-b’ ]diquinoxaline system 5 . Heating in a weaker acid or neat furnished 1-(quinoxalin-2-yl)quinoxalin-2(1 H )-one 12 . The structures of these products were confirmed by independent synthesis and NMR experiments or X-ray crystallography.


Introduction
Treatment of azoxybenzene 1 and its derivatives with certain strong acids is known to result in the corresponding hydroxyazobenzene 2 (Scheme 1). 1 This rearrangement, discovered by Wallach, was named after him. 2 The products of the Wallach transformation have been found to depend on the reaction conditions: the hydroxyl group generally appears in a para position, though the application of photochemical 3 or Lewis acid-catalysed 4 reactions or blocking of both para positions 5 leads to the formation of ortho-hydroxy derivatives.Kinetic studies have resulted in much mechanistic information being deduced from the structural changes in the azoxybenzene 6 and azoxynaphthalene series, 7 but extension of such studies to the heterocyclic azoxy compounds has not been systematically reported.Only the phenylazoxypyridines and their N-oxides were investigated by Buncel and his coworkers. 8We set out to extend the generic Wallach rearrangement to heterobicyclic ring systems, and started our investigations with azoxyquinoxaline 3; 9 this revealed some interesting and surprising reactions and products, depending on the reaction media.This short paper reports our findings.

Results and Discussion
We first applied the original Wallach rearrangement conditions, treating azoxy compound 3 with conc.sulfuric acid in the expectation of obtaining the corresponding 3-oxo (ortho-like product, 4a) or 6-hydroxy (para-like product, 4b) azo compound (Scheme 2).Buncel and his coworkers reported that the phenylazoxypyridines and their N-oxides react much more slowly than does azoxybenzene itself, presumably because of the extra positive charge present in the substrates. 8We therefore decided to carry out the transformation of 3 at higher temperature.After azoxyquinoxaline 3 had been stirred in conc.sulfuric acid at 140 °C for 10 min, the isolated product was recrystallized and characterized by HRMS assay.Surprisingly, we observed nitrous gas evolution and, consistently, HRMS assay did not contain any O atom: instead of the Wallach rearrangement, formally "HNO" was eliminated from 3. Compound 5, with the molecular formula C16H9N5, was obtained in 63% yield (Scheme 3).On the basis of 2D NMR measurements, a new pentacyclic system, imidazo[1,2-a:4,5b']diquinoxaline 5, is proposed for the structure.
This structure was supported by an independent synthesis starting from 2,3dichloroquinoxaline 6 and 2-aminoquinoxaline 7, subjected to the Buchwald-Hartwig protocol, 10 yielding compound 5 in 43%.The synthesis was also carried out in a two-step reaction.The nucleophilic substitution of compound 6 with amine 7 gave 1-(3'-chloro-2'quinoxalinyl)quinoxalin-2(1H)-imine 8, which underwent the Buchwald-Hartwig cyclization 10 to yield pentacyclic compound 5.  Various strong mineral and organic acids uniformly furnished 5 (Table 1, Entries 1-5).In the presence of sulfuric acid, a reduction of the temperature did not have a significant influence on the nature or quantity of the product, but the necessary reaction time increased considerably, from 10 min to 48 h (Entry 2).
A possible formation of pentacyclic derivative 5 is depicted in Scheme 4. In the protonated form 9, 1,2-aryl migration occurred on the diazo moiety, followed by "HNO" loss to give a di(quinoxalin-2-yl)amino cation 10.Then the pentacyclic skeleton 11 was formed by electrocyclization of 10, and after deprotonation pentacycle 5 was obtained.When the value of pKa was systematically increased, dramatic changes were observed above pKa 3.77.Reaction in boiling acetic acid provided 12 instead of 5 (Entry 6, Scheme 5).The HRMS assay indicated the elimination of N2 from 3 to yield quinoxalinylquinoxalinone 12.Its structure was proved by X-ray crystallography: the relative positions of the two planar quinoxaline rings are characterized by a C12-N11-C3-C2 of a torsion angle of 63.2(2)° (Figure 1). 11  The question arose of the possibility of thermal reaction; the same product was obtained from neutral organic solvents (Entries 7 and 8) and even neat from 3. Accordingly, we carried out thermoanalytical studies (DSC, TG and DTG).At the melting point of 3, an intense exothermic reaction (3  12, H166°C  84.6 kcal/mol) was detected, and the gravimetry demonstrated a relative loss of mass m172°C = 9.3%, which is consistent with N2 elimination (theoretical loss: m  9.3%).When the same sample (now containing quinoxalinylquinoxalinone 12) was further heated to 220 °C, endothermic melting (ΔH220°C = 7.2 kcal/mol) ensued.In solvents, the N2 elimination proceeded even at lower temperatures, indicating a very strong solvent effect (Entries 6-8).This type of thermal transformation does not appear to have been widely described in the literature: only one example of the thermolytic loss of N2 from azoxy compounds is known. 13n the evidence of these studies and the literature data, we propose the mechanistic pathway depicted in Scheme 5.The first step involves ipso-attack by the oxygen of the azoxy moiety of 3 on the positively charged C2 of the more distant quinoxaline ring to furnish spiro derivative 13.This is followed by a [2+2]-cycloreversion of intermediate 13 to give quinoxalinone anion 14 and quinoxaline-3-diazonium ion 15.N2 loss occurred during recombination of cation 15 and anion 14 providing quinoxalinylquinoxalinone 12.
To find support for the proposed mechanism we attempted to trap cationic species by a nucleophile.In view of the scope and limitations of the trapping reaction, we set out to catch cation 15 in morpholine.When 3 was heated in boiling morpholine, 2-(morpholin-4yl)quinoxaline 16 was obtained in good yield (Entry 9, Scheme 6). 14Similar treatment of 12 for 10 min resulted in the formation of < 1% of 16.

Conclusions
In summary, the treatment of azoxy compound 3 with strong acids or thermally led to two different reaction pathways, furnishing pentacyclic system 5 and quinoxalinylquinoxalinone 12.
The structures of the products were supported by detailed NMR analysis, and confirmed by independent synthesis (for 5) and X-ray crystallography (for 12).A possible interpretation of the formations of products 5 and 12 is proposed.

Experimental Section
General.Melting points were determined in open capillary tubes with a Büchi 535 apparatus and are uncorrected.NMR spectra were measured with a Bruker Avance 500, Avance 400 or Avance 200 instrument, mass spectra (GC-MS) with a Shimadzu GCMS-QP2010S instrument, highresolution mass spectra with a Waters LCT Premier XE instrument, and IR spectra with a VERTEX 70 instrument (KBr).

Scheme 4 .
Scheme 4. Proposed mechanism of the acid catalyzed reaction of 3.

12 ISSN 5 .
Scheme 5. Proposed mechanism of the thermal reaction of 3.

Figure 1 .
Figure 1.Structure of compound 12 with the crystallographic atomic numbering.

Table 1 .
Influence of the nature of the acid and temperature on the transformation of 3 a Reaction time 10 min; yield after isolation and recrystallization.b Reaction time 48 h.c At boiling temperature.